I've searched around and have not found any real information on ways to improve the tach accuracy.

My NB2 tach running MS3 Basic seems to be off by around 150-200 rpm, I've changed the tach output mode to "Variable" and set the scaling to where it reads accurate across the entire range (in my case it was about 96.5%).
This seemed to be working fine until I noticed at higher RPM (5000+) the tach needle would move in "ticks", stuttering, instead of smooth as it should. I don't remember this happening immediately after the change, only noticed a few days later doing some pulls.

Has anyone messed around with this? Are we doomed to have garbage tach accuracy?

 

My tach is perfectly accurate. It has nothing to do with your ECU.

 

I figured it was the actual physical tach that had the error. It is also not a fixed offset, the error increases with RPM as verified by changing the scaling to 96.5%. Adjusting the needle obviously does not work.

Is there some way to fix the gauge cluster itself?
Seems like the MS3 variable tach output should work just fine, but the notchy behaviour is unexpected.

 

I hesitate to say it's the tach, since they are fundamentally reliable, but the notchy behavior makes me lean that way. Throw a different cluster in it (swap your odo into the new cluster) and see if the issues go away.

 

Will give it a shot and test with another whole dash to rule it out, thanks.
Just to clarify in case I was not clear, when I leave the tach output set to "Fixed" mode as is normal, the tach is perfectly smooth just like OEM, it only gets notchy when I set it to variable mode.
The tach error is consistent with OEM ECU as well. (~100 rpm @ 7000)

 

The tach is inacccurate. I can state this empirically. Feed it mathmatically correct signals for timing, it can vary almost as much as 400rpm at some points - both low and high.

 

I've seen stock NA tachs off by 900rpm at redline.

 

I'm not 100% sure what the ms3 variable mode does - but I know if it's fed a regular square signal, weird things happen(tm). It more or less needs the correct dwell-like signal to function correctly.

 

this. the ECU is accurate.

 

This was my attempt at trying a line of best fit to mathematically correct the tach output on an NB cluster. Instead of trying this really nasty equation, I went with a piecewise function.

 

Seems a bit complex to implement. In reality I'm mostly concerned about accuracy near redline.

As for error in NAs, I can also confirm that I know of many local NAs with errors similar to what you report.

I have used a hand tach to measure crank rpm at the pulley and it matches what the ECU is displaying.
Tomorrow I will play around with variable tach output, maybe the specific value I picked made it misbehave, will test some others.

 

Tested today again with 96.5 and some other values (97, 95 etc..). Made a video to show the issue more clearly (had to use youtube stabilization).

As you can(t) see its perfectly smooth to the eye until ~4000rpm and gets progressively more notchy as it increases. By 6-7k you can clearly see the notchy ticking action.

 

I don't know about the NB clusters, but the NA cluster has an adjustment you can use to correct the tach. It is buried behind the tach and cannot be easily accessed without drilling a hole in the white plastic behind the gauge face. I was able to use Tuner Studio to hold an indicated 5,000rpms on TS and adjust this **** to get the tach to read 5,000 or very close to it. The gauge in my '90 racecar was off by a ton at redline (~850rpms) and using this method I was able to reduce that down to within 100rpms.

I will try and find a picture of this adjustment screw on the web.

 

very informative.

 

Great information. This thread is already totally worth it for future readers

Why drill hole though, can't one just take the dash apart in the same way one does to linearize the water temp gauge?

 

The adjuster is located behind the gauge face pointing towards the turn signal bulbs. It is meant to be used to calibrate the tach on a bench, not in a car. I found it easiest to drill the hole in the white plastic to allow me to adjust it while the cluster was in the car and running. I have not been able to find a good picture of the adjuster screw, and my cluster hood has been on there long enough that taking it off to take pictures means I will be buying a new one.

 

The trimpot is part of an RC circuit that drives the tach needle. It "scales" the tach - so similarly, you can tune it to be accurate at one RPM range at the cost of accuracy everywhere else. This is a NA tach. Trimpot is labelled VR100K at 1 o'clock position.


I took the cover off my NB cluster - no trimpots visible on the driver circuit that's at the bottom.

 

Many thanks for the effort. Shame that the NB one does not have it. Then again, they only seem to be reading ~100rpm high so I can live with it. ( guess I can reset my needle too)